Life insurance is an important household insurance and financial planning product, nevertheless its penetration rate in Iran is very low. One reason is certainly the surrender of insurance policies. However, among other reasons, the focus of the research project was on social factors. Since the latter is quite vast per se, some specific social variables were chosen to serve the general purpose of the study.
Carried out by Dr. Abbas Khandan, Faculty Member of Kharazmi University, the research project is published in 5 chapters including research generals and statement of the problem, literature review and study background, methodology and research data, research findings, and conclusion. The purpose of the study was to investigate the effect of individual characteristics and insurance contract features that affect surrendering term life insurance policies. Given that, a database of 35171 insurance policies issued by a sample company during 2021-2022 was chosen. The policies were purchased in 2016 and 2.25% of them were surrendered up to 2022. In order to discover the reasons, a vast amount of data was collected in order to analyze the behaviors of the customers who surrendered their policies. In this regard, deep learning and neural network algorithms which have very high accuracy in prediction were used as data mining techniques.
The results indicated that from the demographic components, variables including age, gender (female), health surcharge, accident risk rate and from among the contract features, variables such as the term of the insurance policy, the elapsed time since the start date, the premium payment methods with longer installments, higher annual increase of the benefits, and lower the number of risks covered and lower the benefit were negatively correlated with the probability of policy surrender. The results also showed that if the insured is the policyholder himself/herself, the probability of surrender is at its minimum whereas if the insured is someone else, the probability of surrender is higher and it increases for distant relatives.
Those interested to access and download the Persian full text with English abstract of the aforementioned research report, please take a visit to http://www.irc.ac.ir/fa-IR/Irc/4944/Articles/view/16040/1616.